Weaver

The Open-Source NotebookLM Alternative with Infinite Canvas
A self-hosted Google NotebookLM replacement that gives you complete control.
Chat with documents, generate mindmaps, and organize research visually.

Try Live Demo · Quick Start · Features · 中文文档

---

Why Weaver? (NotebookLM vs Weaver)

Tired of Google NotebookLM's limitations? Weaver is a free, open-source NotebookLM alternative that you can self-host and customize.

Feature	Google NotebookLM	Weaver
Open Source	Proprietary	AGPL-3.0
Self-Hosted	Google Cloud only	Deploy anywhere
Visual Canvas	List-based UI	Infinite canvas workspace
LLM Choice	Gemini only	Any LLM (Claude, GPT-4, Gemini, Ollama)
Video Sources	YouTube only	YouTube + Bilibili + Douyin
Data Privacy	Google servers	Your data, your servers
API Access	None	Full REST API
Cost	Usage limits	Pay only for LLM API

Weaver = NotebookLM + Miro + Obsidian — All in one open-source package.

---

Screenshots

Dashboard — Manage all your research projects	Studio — Infinite canvas with Sources & Concepts sidebar
Weaver AI — Chat with your documents, get cited answers	Import Sources — Upload files, paste URLs, or search the web

---

Features

Multi-Source Import (Like NotebookLM, But Better)

Import from multiple sources and let AI help you understand:

• PDF Documents — OCR support for scanned documents (via Gemini or Docling)
• Web Pages — Extract content from any URL
• YouTube Videos — Auto-transcribe with timestamps
• Bilibili & Douyin — Chinese video platform support (NotebookLM can't do this!)
• File Upload — DOCX, CSV, TXT, JPG, and more

Infinite Canvas Workspace (Unique to Weaver)

Unlike NotebookLM's rigid list view, Weaver gives you a visual thinking space:

• Drag-and-drop nodes on an infinite canvas
• Connect ideas with relationship lines
• Rich node types — Notes, sources, AI-generated content
• Canvas tools — Select, pan, link, group, and AI-assist modes
• Zoom & navigate — Smooth zoom with minimap overview

AI-Powered Research Tools

• RAG Chat — Ask questions about your documents with inline citations
• Mindmaps — Auto-generate structured mindmaps from your sources
• Summaries — AI summaries with source references
• Flashcards — Generate study cards automatically
• Articles — Synthesize long-form content from multiple sources
• Concepts — Auto-extract key concepts and relationships

Built for Developers

• Full API access — RESTful API with Swagger docs at /docs
• Bring your own LLM — OpenRouter, OpenAI, Anthropic, or local Ollama
• Extensible architecture — Clean separation of API, domain, and infrastructure layers
• WebSocket support — Real-time document processing status updates

---

Quick Start

Get Weaver running in 5 minutes:

1. Clone & Install

git clone https://github.com/guangtouwangba/weaver.git
cd weaver

# Setup everything (installs uv, system deps, backend & frontend)
# This also creates .env from env.example
make setup

2. Start Infrastructure

# Start PostgreSQL, Redis, and Qdrant
cd app/backend && docker compose up -d

3. Add Your API Key

# Edit the root .env file — only one thing is required:
# Set OPENROUTER_API_KEY (get one at https://openrouter.ai/keys)

Auth is bypassed by default (AUTH_BYPASS_ENABLED=true) so you can start immediately. See env.example for all available options.

4. Run the Application

# Terminal 1: Start backend API (port 8000)
make run-backend

# Terminal 2: Start frontend (port 3000)
make run-frontend

Open http://localhost:3000 and start researching!

See Backend README for detailed configuration and troubleshooting.

---

Architecture

weaver/
├── app/
│   ├── backend/                 # FastAPI + Python 3.11+
│   │   ├── src/research_agent/
│   │   │   ├── api/             # REST endpoints & WebSocket
│   │   │   ├── application/     # Business logic, RAG graphs, AI agents
│   │   │   ├── domain/          # Domain entities & models
│   │   │   ├── infrastructure/  # DB, LLM, vector store, extractors
│   │   │   └── worker/          # Async task queue (ARQ)
│   │   └── docker-compose.yml   # PostgreSQL, Redis, Qdrant
│   │
│   └── frontend/                # Next.js 16 + React 19
│       └── src/
│           ├── app/             # Pages (dashboard, studio, inbox)
│           ├── components/      # UI components (canvas, chat, panels)
│           ├── stores/          # Zustand state management
│           └── lib/             # API client, canvas utilities
│
└── docs/                        # Documentation & design plans

---

Tech Stack

Layer	Technologies
Frontend	Next.js 16, React 19, TypeScript, Konva.js (Canvas), Zustand
Backend	Python 3.11+, FastAPI, SQLAlchemy 2.0 (async), LangGraph
Database	PostgreSQL + pgvector, Qdrant (Vector Search), Redis (Cache & Queue)
AI/LLM	OpenRouter, OpenAI, Anthropic, Google Gemini, Ollama
Embedding	OpenAI, Gemini, Cohere, Voyage
Document Processing	Unstructured, PyMuPDF, Docling (OCR), Trafilatura
Video	yt-dlp, youtube-transcript-api, bilibili-api-python

---

Supported LLM Providers

Provider	Models	Setup
OpenRouter (Recommended)	Any model (Gemini, Claude, GPT-4, Llama, etc.)	OPENROUTER_API_KEY
OpenAI	GPT-4o, GPT-4, GPT-3.5	OPENAI_API_KEY
Anthropic	Claude 3.5, Claude 3	ANTHROPIC_API_KEY
Google	Gemini 2.5 Flash, Gemini Pro	GOOGLE_API_KEY
Ollama (Local)	Llama, Mistral, Phi, etc.	OLLAMA_BASE_URL

---

Use Cases

Students & Researchers

• Upload lecture PDFs and papers
• Ask questions across all your sources
• Generate study flashcards automatically
• Create visual mindmaps for exam prep

Content Creators

• Research topics from multiple sources
• Generate article outlines
• Organize ideas on the canvas
• Export structured content

Professionals

• Analyze reports and documents
• Build a personal knowledge base
• Self-host for complete data privacy
• Integrate with existing workflows via API

---

Prerequisites

• Python 3.11+ / Node.js 18+
• Docker & Docker Compose (for PostgreSQL, Redis, Qdrant)
• macOS: brew install poppler ffmpeg
• Linux: apt install poppler-utils ffmpeg

---

Development

make install          # Install all dependencies
make dev              # Run backend + frontend in parallel
make test             # Run test suites
make lint             # Lint code (Ruff + ESLint)
make migrate          # Run database migrations

API documentation is available at http://localhost:8000/docs when the backend is running.

---

Roadmap

• PDF / Web / YouTube / Bilibili / Douyin import
• Infinite canvas workspace
• RAG chat with citations
• Mindmap generation
• Concept extraction & knowledge graph
• Guest mode for quick testing
• Podcast generation (like NotebookLM Audio Overviews)
• Chrome extension for web clipping
• Multi-language UI
• Collaborative editing

---

Star History

---

Contributing

We welcome contributions! Whether it's:

• Bug reports
• Feature requests
• Documentation
• Pull requests

See CONTRIBUTING.md to get started.

---

Community

• Star this repo if you find it useful!
• Report bugs
• Request features

---

License

Dual Licensed:

• Open Source: AGPL-3.0 — Free to use, modifications must be open-sourced
• Commercial: Contact 819110812@qq.com for closed-source licensing

---

Weaver — The open-source NotebookLM alternative
_{Built with care by the Weaver community
Inspired by Google NotebookLM}

If you're looking for an open-source NotebookLM replacement, give Weaver a star!